Pinball Switch Matrix Not Working: Dead Rows and Phantom Switches

Three drop targets score nothing. The left outlane is dead. None of it is random — those switches sit next to each other in a grid you cannot see, and something they share has failed.

This is the switch matrix. Almost everything written about it online is aimed at 1990s WPC machines, which is no help when your game is a 1979 Bally or a 1983 Gottlieb: the wiring idea is the same, but the chips, connectors, test mode and even the correct diode are not.

The short version: read the pattern before you touch anything. One dead switch is a switch or its own diode. A whole group dead is a shared wire, a connector pin or the chip driving that line. Switches reading closed when they are not is almost always a shorted or reversed diode. Work out which of the three you have and the job shrinks from forty switches to about three.

Contents

What the switch matrix actually is

A pinball machine has dozens of switches and a processor with nowhere near dozens of spare inputs. The answer the industry settled on around 1977 was a grid.

Switches are wired as a matrix of strobe lines (columns) and return lines (rows). The CPU energises one strobe at a time, very fast, and watches which returns come back. A switch where strobe 3 crosses return 5 is uniquely identified by that pair — five strobes and eight returns give forty switch positions from thirteen wires.

Each switch needs a diode in series with it, a one-way valve. Without it, closing two switches creates a sneak path through a third and the CPU reads a switch nobody touched. The banded end marks the direction it blocks. That small component is the most common failure in the whole system. For the wider map this sits inside, see our guide to pinball machine electronics.

Read the pattern first

This is the step people skip and the one that saves the afternoon. Put the game in switch test, close every switch by hand, and write down what registers. The shape of the failure tells you where to look.

What you see What it usually means Where to look
One switch does nothing Mechanical, or its own diode is open Contacts, solder joints, that diode
A whole row or column dead A shared wire, connector pin or driver That line end to end
One switch closes several A shorted or backwards diode The diode on the switch you pressed
A switch reads closed and isn't Stuck switch, shorted diode, lead touching That switch, then its neighbours
Faults move as you play Connectors, or grounding Backbox connectors, then grounds

Two failing switches that are physically far apart on the playfield is the giveaway. Find them in the matrix chart in your cabinet schematic and see which strobe or return they share. That line is your fault.

Getting into switch test

Every platform does this differently, which is part of why generic advice fails.

Bally and Stern, 1977–1985. Step into the self-test with the red button inside the coin door. First: remove the ball, reset all drop targets up, and check that spinners hang straight down — a spinner resting half open reads as closed and will send you chasing a fault you created. With the playfield quiet the display should show zero; a number instead is a switch the game thinks is closed. Note the quirk that catches everyone: only the lowest-numbered stuck switch appears. Fix it, power up, and a second may show — not a new fault, just one that was hidden.

Williams System 3 to 7. The diagnostic sequence steps through lamps, displays, solenoids and switches in turn. If the game is not reaching test mode cleanly, start with the CPU LED instead — our guide to Williams System 7 diagnostic codes covers what it is telling you.

Gottlieb System 80, 80A and 80B. The switch test reports closures on the displays as Cx-y, which you translate using the matrix chart in the game's manual. One warning that has wasted many evenings: from the later System 80B games, Gottlieb stopped printing some cabinet switches in that chart. The advance, replay, play/test, coin and tilt switches are real matrix members that several titles simply do not list. A position registering when the chart says it does not exist is usually one of those, not a fault — the cabinet schematic still shows them.

The diode is the part that fails

There are two ways a switch diode goes wrong, with opposite symptoms.

  • Open. Current cannot flow at all, so the switch never registers however well the contacts close. It looks exactly like a dead switch.
  • Shorted or fitted backwards. The valve no longer blocks, current sneaks the wrong way, and closing one switch lights up others on the same lines. This is the "phantom" switch — the classic cause of a game that scores things nobody hit.

A diode lead bent over and touching an adjacent lug does the same as a short, and is common on a machine that has been worked on before. Look before you unsolder.

The Gottlieb diode trap

Here is the detail that catches careful people, because the usual advice is wrong for these machines.

Gottlieb used germanium diodes — typically the 1N270 — in the switch matrix, chosen for their very low forward voltage of around 0.2 V, which suits the logic levels involved. On a meter's diode range you will read roughly 0.2 to 0.3 V, not the 0.6 to 0.7 V silicon gives.

Two consequences. Expecting 0.6 V, you will condemn a good diode. And replacing one with the 1N4148 every parts drawer contains roughly triples the drop on that switch — it will often still work, but it is exactly the marginal change that produces an intermittent nobody can find later. For a Gottlieb matrix diode, a small-signal Schottky such as a BAT42, BAT43, BAT46 or BAT85 is the sensible modern substitute, keeping the low forward voltage the design assumed.

Bally, Stern and Williams matrices are conventional silicon and are content with a 1N4148, interchangeable with the older 1N914. Know which machine you are in front of before you reach for the drawer.

Testing a switch and its diode

You need a multimeter with a diode range. Machine off, and unplug the relevant switch connector from the board so you are measuring the switch rather than the whole game.

  1. Set the meter to diode. It reads a forward voltage drop, not resistance — this trips people up.
  2. Black probe on the banded end, red on the other, switch held closed.
  3. Expect roughly 0.2–0.3 V on Gottlieb germanium, 0.6–0.7 V on silicon.
  4. Reverse the probes. Now expect nothing. A reading in both directions means a shorted diode.
  5. Nothing either way, with the switch definitely closed, means an open diode or an open switch — bridge the contacts with a clip lead to tell them apart.

For a whole dead row or column, skip the diodes and test continuity instead: from the switch lug on the playfield back to the pin at the board connector. A break in that run is a broken wire or a bad crimp, found without removing a single component. Dirty and pitted contacts also cause plenty of "dead" switches that are not electrical faults at all — our maintenance guide covers cleaning them.

Where the matrix lives, platform by platform

Bally and Stern MPU, 1977–1985

A 5 column by 8 row matrix — forty positions — running through a single 6820/6821 PIA at U10 on the MPU board: the PA side drives the strobes, the PB side reads the returns. At connector J2, pins 1–5 are the columns and pins 8–15 the rows, which makes a dead line easy to trace to a pin.

Two Bally quirks. The coin door switches have no diodes — Bally left them off — so a mangled coin switch throws errors all over the game. And small disc capacitors (0.050 µF originally, 0.047 µF today) sit across some switches for noise suppression; a shorted one behaves exactly like a stuck switch. This is the same across the range, from Evel Knievel and Eight Ball onward. If the board is not booting cleanly first, start with what the Bally MPU LED flashes mean.

Williams System 3 to 7

Williams put things where people do not expect. Through System 6 the machine uses four PIA chips: one on the MPU board for the displays, and three on the driver board — solenoids, lamps, and one that reads the switches. So a dead switch column here often means a driver board fault, not a CPU fault. Checking the CPU first is the natural instinct and the wrong one.

The other Williams truth is blunter: the 40-pin interboard connector causes the overwhelming majority of intermittent faults on these games, through hairline cracks in forty-year-old solder joints. Before suspecting anything clever, unplug and reseat the boards — the wiping action alone cures a surprising number of faults. If it returns, the header pins want resoldering. Connector IJ2 and its single +12 V wire is a known cause of resets during heavy play. All of which applies to Gorgar and the rest of the range.

Gottlieb System 80, 80A and 80B

The System 80 CPU examines 64 switch positions. Returns arrive at input port A of the 6532 RIOT at U4; U5 decides whether the CPU reads the external playfield matrix or the on-board DIP switches. The strobes leave through a 74154 decoder at Z33 — worth knowing, because a 74154 with one dead output kills an entire strobe line and everything on it, presenting exactly as "a whole group of switches stopped at once".

System 80 also has a grounding scheme that generates an unreasonable number of faults looking like anything but grounding. If your switch problems wander, read our System 80 ground mods guide before replacing a chip. For one stubborn target switch on a game like Black Hole, we have walked through diagnosing a dead target switch on a System 80 step by step.

Gottlieb System 1

System 1 is the awkward one. Switch input and output run through U5, the A1752CF, a custom Gottlieb part that is effectively unobtainable with no drop-in equivalent. It is also fragile: static discharge and stray voltage from the solenoid bus can kill it, and the board has no clamping diodes on the switch input resistors to prevent that. Adding protection diodes there is a well-known modification worth doing on a machine you mean to keep, such as a Cleopatra. Note the distinction that confuses people: the series diode at each switch is a different component from the clamping diodes protecting the CPU inputs. System 1 has the first and lacks the second.

When the fault really is on the board

Most switch matrix faults are not on the board. They are a diode, a broken wire, a bent contact or a green connector pin — a few euros and an evening. We would rather say that plainly than sell you something you do not need.

It changes when the board is what is taking the matrix out. Chiefly battery corrosion, where leaked electrolyte creeps along traces and eats switch return lines like anything else — a board can be cleaned, but corrosion hides in vias and under ICs where you can neither see nor reach (our guide to spotting and stopping battery corrosion covers what is recoverable). Or a failed chip you cannot buy: a dead 74154 is a two-euro part, a dead A1752CF is not a shopping trip.

BallyFA replacement MPU board for Bally and Stern pinball machines 1977-1985, the board where the switch matrix strobe and return lines terminate at connector J2
On a Bally or Stern machine the whole switch matrix terminates on the MPU board at connector J2 — columns on pins 1–5, rows on pins 8–15. The BallyFA is a modern replacement for that board. Photo: Pinballs Store.

Where replacement makes sense, the BallyFA covers the Bally and Stern MPU of this era (our Bally MPU buying guide sets out the options), and the WillFA7 replaces both MPU and driver board on Williams System 3 to 7 — which matters here, because the Williams switch PIA lives on the driver board. On Gottlieb System 80, a Lisy80 lets you watch switch closures live in a web interface instead of squinting at the displays. It is built on the LISY project, an independent developer's free, non-commercial DIY project — we supply it ready to use and support it afterwards; we did not create it.

We test boards before they ship. Either way, keep your original: a matrix fault rarely kills a board outright, and it makes a good winter project once the machine plays again. And if your fault turns out to be a bent switch blade, that is a good outcome, not a wasted evening.

FAQ

Why is a whole row of switches not working on my pinball machine?

Because those switches share a wire. A whole row or column dying at once points to the shared return or strobe line rather than the switches: a broken wire on the playfield, a bad pin at a board connector, or the chip driving that line. Test continuity from a switch lug back to the board connector pin before suspecting the board.

What does it mean when the switch test shows a switch closed that isn't?

Usually a shorted or backwards diode on that switch, or a diode lead touching another lug. It can also be genuinely stuck — a spinner hanging half open or a drop target that has not fully reset both read as closed. On Bally machines only the lowest-numbered stuck switch is displayed, so a second may appear once you fix the first.

Can I use a 1N4148 for a pinball switch matrix diode?

On Bally, Stern and Williams, yes — 1N4148 and the older 1N914 are interchangeable and correct. On Gottlieb it is not ideal: Gottlieb used germanium diodes such as the 1N270, with a much lower forward voltage. A 1N4148 will often work but roughly triples the drop on that switch. A Schottky such as a BAT42, BAT43 or BAT85 is the better modern substitute.

How do I test a pinball switch diode with a multimeter?

Machine off, switch connector unplugged from the board, meter on its diode range rather than resistance. With the switch held closed you should read a forward voltage drop one way — about 0.2–0.3 V on Gottlieb germanium, 0.6–0.7 V on silicon — and nothing the other way. A reading in both directions means a shorted diode; nothing in either means an open diode or an open switch.

Why do two switches trigger together when I press one?

The classic sneak-path symptom. Without a working one-way diode, current from the switch you pressed finds a route back through other switches on the shared lines and the CPU reads closures that never happened. Check the diode on the switch you pressed: shorted, reversed, or its lead touching something it should not.

My Gottlieb switch test shows a position the manual doesn't list. Is the board faulty?

Probably not. From the later System 80B games, Gottlieb stopped printing several cabinet switches in the manual's matrix chart, including the advance, replay, play/test, coin and tilt switches. They are still real positions. Check the cabinet schematic, which shows them even when the chart does not.

Sources

Stuck on a matrix fault? Send us the game, which switches misbehave and what your meter reads across the diode. Our technical support is free, before and after any purchase — including when the answer turns out to be a spinner hanging half open.

Is your machine down right now?

Tell us your machine and the exact symptom — we'll tell you which board you need, or whether you need one at all. It's free, and Valère answers personally.

Describe my fault →

Not ready to order yet?

Get our troubleshooting guides, restoration tips and new product news — about one email a month, never spam.

Back to blog

Nos produits

View all